2

On rank and successor relations: numerical and non-numerical expression

of relative position in Akan Clement Kwamina Insaidoo Appah, University of Ghana

There are two principal means of expressing relative position/rank in ordered

sequences – numerical and non-numerical means. For Akan, Christaller (1875)

subsumed both of them under the heading of ordinal numerals. Based on data drawn

from a variety of sources, we attempt to make a clear distinction between the two in this

paper, noting that a numeral must express the properties of empirical object by means

of numbers. We find this to be true of only one class of the constructions. Therefore,

they are qualified as ordinal numerals and they can refer to the exact ranks of ordered

items numerically. The other construction type, which refers to successor relations non-

numerically, are clearly identified and their properties discussed. It is shown that the

two kinds of constructions divide any sequence of ordered items, called the ordinal

space, differently. Ordinal numerals partition the ordinal space into two, first position

and others, excluding the last, which are referred to specifically by means of cardinal

numeral constituents. The successor relation constructions divide the ordinal space into

three (first, last and others in-between) or two (first and next) where the speaker

deliberately avoids mentioning the expression for the last position in a set of ordered

items. Ideas and formalism from Construction Morphology are employed for the

analysis and presentation of the data.

Keywords: Akan, Construction Morphology, Constructional Idiom, Ordinal numeral,

Rank, Schema, Successor relation

1. Introduction

This paper deals with how rank within ordered items and successon between items in a

sequence are expressed in Akan. It shows that the expression of rank and succession are done

numerically and non-numerically respectively and that the constructions employed tend to

partition any sequence of ordered items differently. The basis of the distinction is number

assignment, which refers to the use of number to assess properties of empirical objects in

different contexts, so that relations between numbers are associated with relations between

empirical objects (Dehaene 1997, 2001a, 2001b; Wiese 2003a).

Three basic types of number assignments are distinguished – cardinal, nominal and

ordinal number assignments (Wiese 2007: 759-760). Of the three, the one that is specifically

relevant to the focus of this paper is Ordinal number assignment (e.g., set 4) which associates

the ordering relation in a number sequence (‘<’ or ‘>’) with the relative ranks of objects in an

empirical sequence. For example, as Wiese (2003a) observes, in relation to the ranks of runners

in a race, the relation ‘>’ is associated with ‘finish faster than’, so that if person A finishes as

the sixth runner and person B comes in as the eighth runner, then “A > B” means A was faster

than B. In cardinal number assignment (e.g. five houses), the empirical relation ‘has more

elements than’ (represented by the numerical relation ‘>’), expresses a relation between sets,

so that the more elements a set has, the higher the number it receives. In this way, positions in

3

the number sequence can be said to identify the size or cardinality of empirical sets.1 In nominal

number assignment, the numeral relation ‘=’ (or ‘≠’) is associated with the empirical relation

‘is identical (or non-identical) with’. This is how numbers come to identify elements within a

set, making the number a label (Wiese 2003a). An example is how a particular route that is

plied by a bus comes to be known simply by the bus number.

Consistent with the observation that “[n]ot all languages have a separate series of

ordinal numerals” (Hurford 2000:71), Christaller (1875: 54) observed that the kind of ordinal

numerals in European languages which denote the place that any item holds in a series, do not

exist in Akan. In their place, various types of verb phrases (VPs) are employed. Christaller then

identifies several exemplar VPs which he calls ordinal numerals. He observes that the first

position is expressed as di kan (1). It is formed from the verb di, which has many meanings of

which the relevant one is “to occupy”, and its complement kan which is a noun denoting “the

first or foremost (or former) place or time in a series of places or events”.

(1) di kan

occupy front

‘to occupy front/be first’

The second person in a series of people, according Christaller, is rendered as nea odi hɔ (2a),

as found in odi hɔ, lit. ‘he occupies the (next) place there’; the second thing in a series of things

is rendered as nea edi hɔ (2b). Also, for the persons in second, third, fourth, etc., positions

Christaller provides, respectively, the constructions nea otia (2c) or ɔto so abien, abiesa, anan,

etc., (2d). The meaning of the verb tia is ‘to add (in order to fill up or make up a sum)’, while

the meaning of the phrase tɔ so is rendered as ‘to lay (or lie) above or upon’.

(2) a. nea2 o-di hɔ

person.who 3SG-occupy there

‘person who occupies the next place there/the second person’

b. nea e-di hɔ

thing.which 3SG-occupy there

‘thing which occupies the next place there/the next thing’

c. nea o-tia abien

person.who 3SG-add up to make two

‘the second person’

d. nea ɔ-to so abien

person.who 3SG-lie upon two

1 A common way of verifying cardinal number assignment is counting, which establishes a one-to-one mapping

between the elements of a set and an initial sequence of natural numbers. Thus, we end up using exactly as many

numbers as there are objects so that the counted set and the set of numbers used in the count have the same

cardinality. Wiese (2003a: 385) argues that “[b]ecause the numbers form a fixed sequence, we always end up with

the same number for sets of the same cardinality. Hence, this number can be used to identify the cardinality of a

set, … owing to its position in the number sequence”. 2 According to Saah (2010), the form nea results from the fusion of the noun which is modified by the relative

clause (like nipa ‘person’/ade ‘thing’) and the relativizer a.

4

‘lie upon two/be second’

For the final position, Christaller provides constructions like odi akyiri ‘he occupies the back-

part’ (3a), ɔka akyiri ‘he remains behind’ (3b), otwa to ‘he cuts off the hind end’ (3c), etc.

(3) a. o-di akyiri

3SG-occupy back

‘S/he occupies the backpart’

b. ɔ-ka akyiri

3SG-remain back

‘S/he remains behind’

c. ɔ-twa to

3SG-cut.HAB hind

‘S/he cuts off the hind end’

The structure of these constructions is explained further bellow, in sections 5 and 6, where it is

shown that there may be multiple means of expressing the same ordinal meaning in Akan.

Indeed, the presence of multiples expressions in a language for positions in ordered sequences

is well attested cross-linguistically (see, for example, Mel'čuk 1994, Veselinova 1997, Stump

2010, Stolz & Veselinova 2013). Thus, Christaller’s presentation on ordinal numerals in Akan

is generally consistent with prevailing approaches to delineating what counts as an ordinal

numeral, where all kinds of constructions, including those that do not contain number words,

are regarded as ordinal numerals to the extent that they refer to relative ranks of items in ordered

sets, such that there is no distinction between numerical and non-numerical reference to ordered

items (cf. Corbett 1978, Hurford 1987 and Stolz & Veselinova 2013). However, a close look

at the set of constructions that Christaller calls ordinal numerals reveals that we can distinguish

between constructions that express relative rank by means of number words and those that

express succession without the use of number words. The former remains called ordinal

numerals and the latter, successor relation constructions or succession constructions. There are objections to the grounds for the proposed distinction. It has been argued that

the absence of a number word is not a sufficient criterion for a linguistic unit not to count as a

numeral. We share this position too. However, fundamental to the distinction made in the

present paper is the notion of number assignment which associates the relationship between

numbers with the relation between empirical objects, described as the use of number to assess

properties of empirical objects. For ordinal expressions, this property is the rank of objects in

a sequence (Wiese 2003a, 2003b, 2007; Appah 2019). Therefore, whatever construct is

classified as an ordinal numeral or is purported to be used for ordinal number assignment must

be a numeral, defined as a linguistic expression (word or phrase) with a primary function of

expressing a unique number, and must identify a specific property of an empirical object, either

collection or sequence (Veselinova 1997: 445).

Thus, for the purpose of this paper, an ordinal numeral is a numeral which identifies a

specific rank within a sequence, much in agreement with the observation that “[a] numeral is

used with ORDINAL MEANING when applied to an individual object in an ordered sequence,

often in connection with a singular noun” (Hurford 1987:168). That is where the class of

successor relation constructions fail to qualify as ordinal numerals in Akan. They simply

5

express non-specific successor relations within ordered sequences. They are not primarily

numerals in that they are not dedicated to the expression of unique numbers and there is no

evidence that, in their current form, they are on the lexicalization/grammaticalization path to

becoming numerals. Hence, they can be used for any position at all relative to an already

existing position. Additionally, as discussed in section 6, we may have actual ordinal numerals

formed from them by the addition of number words. The rest of the paper is structured as follows: In section 2, I briefly present the

methodology, outlining where and how the data were obtained. In section 3, I briefly present

Construction Morphology, the theoretical framework from which I employ formalism for the

analysis and presentation of the data. In section 4, I present what Appah (2019) calls the ordinal

space, a characterization of any sequence of ordered items or positions. In section 5, I review

what Appah (2019) presents about Akan ordinal numerals, constructs which express rank

numerically and can refer to unique positions in the ordinal space. I also show another means

of expressing ordinal first which had not been previously reported in the literature. Section 6

deals with successor relation constructions, focusing on a detailed description of their form and

distribution. For each class of constructions in sections 5 and 6, I show how they partition the

ordinal space. Section 7 concludes the paper.

2. Data and methodology

Data for this study were drawn from a variety of sources, including the Akan translations of

the Holy Bible (Fante & Akuapem dialects) and a grammar (Christaller 1875), and were

supplemented by my native speaker intuitions. The data on ordinals that were collected from

Christaller (1875), as shown in (1) – (3) above, formed the primary basis for the analysis, given

that he clearly identified almost all the classes of construction that referred to succession and

rank in Akan. The Akan translations of the Holy Bible were looked up for contexts in which

the numerals occurred, as shown in sections 5 and 6. A principal reason for choosing illustrative

constructions from the Bible was that direct free translations could be found in the English

Bible, New King James Version (NKJV), thus, eliminating any possibility of mistranslation.

Except otherwise stated, all the examples cited come from the Fante dialect of Akan.

However, the analysis and the claims made therein apply to all the dialects of Akan because,

as far as the constructions at issue are concerned, the relevant differences between the dialects

tend to be mainly phonological (cf. Dolphyne 1988). For example, ordinal first, which is

realized as dzi kan in Fante, is rendered as di kan (1) in Akuapem and Asante, while fifth, is

realized as (4a) in Fante, and as (4b) in Akuapem and Asante. The dialectal variation in the

realization of the relator noun do (Fante) versus so (Akuapem/Asante) is purely phonological.

(4) a. tɔ do enum (Fante) b. tɔ so nnum (Asante)

lie on five lie on five

‘lie in position/come in at number five/be 5th’ ‘lie in position five/be 5th’

There could be semantically motivated phonological differences too. For example, the ordinal

numerals are integrated into the grammar of the NP by relativization because they are VPs and

cannot directly modify the head noun. Thus, the VP forms part of the sentence which is

introduced by the relative marker <a>, in (5).

6

(5) a. Nipa a o-di kan no

person REL 3SG-assume front CFD[Clause Final Determiner]

‘the first person’

b. nwoma a e-di kan no

book REL 3SG-assume front CFD

‘the first book’

We observe a difference in the quality of the third person singular (3SG) pronoun between the

animate subject <o-> in (5a) and the inanimate subjects <e-> in (5b). This animacy distinction

is observed in the Asante and Akuapem dialects only.

We will not comment any more on these dialectal differences because they are not

relevant to the specific issues that are discussed in the rest of the paper.

3. Construction Morphology

As noted above, we employ ideas and formalism from Construction Morphology (CxM) in the

analysis and presentation of the data. CxM is a theory of linguistic morphology whose goal is

to present a framework in which the differences and commonalities of sentence-level and word-

level constructs are accounted for adequately and consistently, thus, providing “a better

understanding of the relation between morphology, syntax and lexicon and of the semantic

properties of complex words” (Booij 2010a: 543).

Central to CxM, is the notion construction as developed in Construction Grammar

(Goldberg 1995, 2006; Fillmore et al. 2003; Michaelis & Lambrecht 1996; Jackendoff 1997,

2002, 2008), which is characterised as a pairing of form and meaning, formed by means of

schemas, which are abstractions over sets of actually existing complex forms. Thus, schemas

first express predictable properties of existing constructions and also serve as blueprint for

assembling other constructions of comparable complexity (Booij 2007, 2010b; Appah 2013).

This is shown by the schema in (6) which generalizes over all right-headed compounds.

(6) ⟨ [[a]Xi [b]Yj]Yk ↔ [SEMj with relation R to SEMi]k ⟩ The upper-case variables X and Y stand for the major lexical categories (N, V & A). The lower-case

variable a and b stand for arbitrary strings of sound segments, whilst i, j and k are indexes for the matching

properties of the compound and its constituents.

Schemas and their instantiating constructions co-exist in a hierarchically structured lexicon,

where two types of relations hold – “instantiation”, which exists between a schema and a

construction that is formed by the schema, and “part of”, which obtains between a construction

and its constituents. These are illustrated in (7), where each dominated constructional schema

is an instantiation of the one that dominates it and the individual constituents, school and bus

are ‘part of’ the compound school bus.

A schema in which at least one of the constituents is lexically fixed or prespecified is

called a constructional idiom (Jackendoff 2002; Booij 2002). Here, the form that fills the slot

lexically is deemed to be part of the constructional schema, so that it is only the variable slot

that would be available, on occasion, to be filled to form different instantiations of the

construction. The idea of constructional idiom will be employed in the analysis to show that

7

the general properties of the various classes of constructions discussed in this paper may be

captured straightforwardly in schemas and constructional idioms that abstract over the

properties of the ordinal and succession constructions at issue in this paper.

(7) ⟨ [[a]Xi [b]Yj]Nk ↔ [SEMj with relation R to SEMi]k ⟩

⟨ [[N]i [N]j]Nk ↔ [SEMj meant for SEMi]k ⟩

⟨ [[school]i [bus]j]Nk ↔ [busj meant for schooli]k ⟩

[school]N [bus]N

The constructionist view of the lexicon is what Jurafsky (1992) characterises as the

constructicon, a lexicon populated by all types of constructions. Therefore, constructions can

inherit all kinds of properties from other sanctioning constructions. In this paper we will argue

that Akan ordinal numerals and successor relation constructions do inherit their formal

structure from VPs in Akan.

An important property of constructions is that they are not expected to be

compositional. Rather, they are expected to be predictable, and they can have properties that

do not emanate from the properties of their constituents, called holistic properties (Booij 2010b,

2012; Appah 2013, 2015, 2017). In CxM, therefore, both compositional and extra-

compositional properties of constructions can be accounted for straightforwardly, obviating the

need to posit abstract categories to carry extra-compositional semantic components of

constructions (cf. Jackendoff 1997, 2008, Goldberg & van der Auwera 2012, Appah 2013,

2016, 2017, Lawer 2017, Dugas 2018, De Wit 2018, Broohm 2019).

4. The ordinal space

In Appah (2019), the idea of the ordinal space was introduced. Presented graphically as a line

with various positions marked, as shown in Figure 1, it is meant to be a pre-theoretic

characterisation of any ordered sequence of positions that are, for example, assigned values in

ordinal number assignment (cf. Wiese 2003b). This will be equivalent to Hurford’s context-

given sequence, as contained in the observation that “to interpret an expression containing an

ordinal numeral, a particular ordered sequence of objects must be present, explicitly or

implicitly, in the context in which the expression is used. I will call such a sequence the

'context-given' sequence” (Hurford 1987: 170).

1st 2nd 3rd 4th 5th 6th .....

Figure 1: The ordinal space

As Appah (2019) observes, assigning values to positions in the ordinal space facilitates

reference to the positions. However, reference to a position in the ordinal space must not

necessarily be by means of a number, if the intention is not to be specific, as the discussion

8

below shows. The first position in the ordinal space is always known (marked with eer-ste in

Dutch and first in English). However, if the items in the ordinal space do not constitute a finite

set, then there would be no non-arbitrary way of determining the terminal position in the ordinal

space (Appah 2019). This is because, whilst it is clear that no one ever counts ad infinitum, one

cannot rule out the possibility of someone counting one more than any current terminal number

(cf. von Mengden 2010; Stampe 1976). That notwithstanding, given the possibility of a finite

set, languages make provision for identifying the final position in an ordered set, sometimes

with a dedicated lexical item like last in English.

The next two sections deal with two classes of Akan constructions used for expressing

relative position in the ordinal space. In the discussion, we show graphically how the various

constructions partition the ordinal space.

5. Akan ordinal numeral constructions

The first group of constructions for expressing relative position are ordinal numerals. They are

the default constructions for expressing the rank of items (first, second, third, etc.) in any

ordered sequence (Stump 2010), so that a particular element of a set is assigned a place within

that fixed order (Stampe 1976: 600; von Mengden 2010: 21). In their description of the

functions of ordinal numerals, Stolz and Veselinova (2013), observe that typically ordinal

numerals identify the position that a member of a set occupies relative to other members of the

same set (e.g. the fourth day). They go on to argue that the main functions of ordinal numerals

comprise the identification of ranks within a hierarchy and the temporal order in a sequence of

events or the like. This hierarchy/temporal order is referred to as “context-given sequence”

(Hurford 1987: 170) or the ordinal space (Appah 2019).

As noted above, Akan ordinal numerals take the form of VPs and the first position is

expressed as dzi kan, as shown in (8). It is found occurring in the Akan translation of the names

of the books of the Holy Bible which come in sets, such as the books of Samuel, the first of

which is rendered as shown in (9).

(8) dzi kan

occupy/assume front

‘to be first’ [lit. to occupy the front/to lead]

(9) Samuel nwoma a o-dzi kan no

Samuel book REL 3SG-assume front CFD

‘the first book of Samuel’ (lit. the book of Samuel which leads) (Appah 2019: 6-7)

Clearly, this construction is not a numeral sensu stricto because it neither contains an

actual number word nor is it dedicated solely to the expression of the relevant number.

Therefore, it cannot be said to involve the use of number to express a quantitative property of

an empirical object, as is required of numeral constructions in the context of number

assignment (cf. Wiese 2003a, 2003b, 2007). It is worth noting, however, that Akan is not

unique in employing a construction which is neither a numeral nor contains an actual number

word for ordinal first. Indeed, the idiosyncratic nature of the Akan construction for first is

consistent with the observed nature of expressions for first cross-linguistically (cf. Barbiers

2007; Booij 2009; Stolz & Veselinova 2013).

9

A noteworthy feature of the phrase dzi kan is its meaning; on its own, this construction

simply means ‘to lead’ or ‘to assume the front positions’. This also is consistent with the

meanings of constructions that are employed for the expression of ordinal first. As Veselinova

(1997:430) observes, commonly noted meanings with the tern for first include ‘foremost’,

‘earliest’, ‘best’ and ‘most important/eminent’. Hence, the literal meaning of the construction

in (9), which contains the ordinal first, is ‘the book of Samuel which leads’. This point, coupled

with the fact that the construction does not contain an actual number word, makes it compelling

to argue that the ordinal meaning does not come from the VP dzi kan per se. Rather, it has to

be regarded as a property of the whole construction which is embedded in the larger

construction by means of relativization (cf. Appah 2019). Thus, for the CxM representation,

we posit the schema in (10) with the semantic specification on the right end of the double

arrow, where ORD is a semantic operator with scope over the meaning of the entire

construction. Through co-indexation of the VP with the meaning of the whole construction we

capture the intuition that the ordinal meaning is a holistic property of the construction (cf.

Appah 2019).

(10) < [[dzi]i [kan]j]k]l ↔ [ORD [SEM]k]l > ‘first’

We notice that it may be possible to have the word kan ‘front’ alone expressing first.

This is found in the Akan translation of 2 Kings 1:14, as shown below in the Akuapem (11a)

and Fante (11b) dialects. The context is that the king sent three captains with their troops of fifty

soldiers with the mandate to arrest the prophet Elijah. The first and second captains and their troops had

been consumed by fire, at the command of the prophet (2 Kings 1:9-14). Therefore, fearing for his life,

the captain of the third group came to plead with the prophet to spare their lives, given that fire came

down and burned up the first two captains and their fifty soldiers.

(11) a. Hwɛ, ogya a-fi soro […] a-bɛ-hyew kan

look, fire PERF-come.from above […] PERF-come-burn first

eduonum so asafohene baanu no ne wɔn eduonum […]

fifty over captain two DEF CONJ 3POSS fifty […]

‘Look, fire has come down from heaven and burned up the first two captains of fifties

with their fifties’ (NKJV)

b. Hwɛ, ogya fi sor siane-e bɛ-hye-ew nkan no

look, fire come above descend-PAST come-burn-PAST first DEF

eduonum do asafohem-fo beenu no nye hɔn eduonum […]

fifty over captain-PLU two DEF CONJ 3POSS fifty […]

‘Look, fire came down from heaven and burned up the first two captains of fifties with

their fifties’ (NKJV)

This use of (n)kan alone for ordinal first has not been previously reported in the literature on

Akan numerals. This might be because the word is used adverbially elsewhere in the grammar

with the meaning ‘previously/in the past’. However, as the examples in (11) show, it can be

used for ordinal first, where it occurs before the noun or, in the present case, the numeral that

it modifies. Again, the adverbial meaning of (n-)kan ‘previously’ accords well with the

meanings commonly associated with terms for first, including ‘earliest’ (Veselinova 1997: 430;

10

Greenberg 2000: 780). Thus, what appears to be the lexicalization of a form meaning ‘front’

for ordinal first is consistent with observed cross-linguistic patterns.

Unlike the construction for first which does not contain a number word, beginning from

second, Akan ordinal numerals identify unique positions in the ordered set by means of cardinal

numerals. Thus, any position, between first and last, may be specifically referred to numerically

and the relevant Akan numeral constructions take the form of VPs, as noted by Christaller

(1875) and consistent with the view in Appah (2019) that the VP structure of Akan ordinal

numerals, suggests that Akan numerals inherit their formal structure from already existing

structures in the language.3

Two subtypes of VP-ordinal numerals are identified. The first is built around the verb

tsia [ʦĩã] ‘to pile/to add (in order to fill up or make up a sum)’ and a cardinal numeral that

identifies the relative position in the ordinal space, resulting in [V NUM]VP, as shown in (12).

(12) a. tsia ebien b. tsia du

pile.on two pile.on ten

‘added to make two (2nd)’ ‘added to make ten (10th)’

c. tsia anan d. tsia awɔtwe

pile.on four pile.on eight

‘added to make four (4th)’ ‘added to make eight (8th)’

As noted in Appah (2019), this class of ordinals occur in the Akan translation of the names of

certain books of the Bible, such as the Pentateuch, the first five books of the Bible which are

attributed to Moses. The last four of the books are named as shown in (13).

(13) a. Moses nwoma a o-tsia ebien a wɔ-frɛ no Exodus No

Moses book REL 3SG-piles.to.make two REL 3PL-call 3SG Exodus CFD

‘The second book of Moses which is called Exodus’

b. Moses nwoma a o-tsia ebiasa a wɔ-frɛ no Leviticus No

Moses book REL 3SG-piles.to.make three REL 3PL-call 3SG Leviticus CFD

‘The third book of Moses which is called Leviticus’

c. Moses nwoma a o-tsia anan a wɔ-frɛ no Nkanee

Moses book REL 3SG-piles.to.make four REL 3PL-call 3SG counting

‘The fourth book of Moses which is called Numbers’

d. Moses nwoma a o-tsia enum a wɔ-frɛ no Deuteronomy No

Moses book REL 3SG-piles.to.make five REL 3PL-call 3SG Deuteronomy CFD

‘The fifth book of Moses which is called Deuteronomy’

As noted above, these constructions are integrated into the grammar of the noun phrase by

means of relativization. This is because, being VPs, they cannot be direct modifiers of the head

noun (cf. Appah 2019).

3 This may be true of numeral constructions in other languages of the world also (cf. van Katwijk 1965; Brainerd

1966, 1968; Brandt Corstius 1968; Corbett 1978; von Mengden 2010).

11

The constructional schema for the group of ordinal numerals that are built around the

verb tsia is a subschema of Akan transitive VPs, a constructional idiom, in which the first

constituent is pre-specified as tsia and the second constituent is required to be a cardinal

numeral, as shown in (14).

(14) < [[[tsia]i [NUM]j]k]l ↔ [ ORD [SEM]k]l >

Like the construction for first, the ordinal semantic operator in (14) has scope over the meaning

of the entire construction which contains the cardinal numeral. This makes the semantic

operator in question function differently from the ordinal suffix (-th) found in the English

ordinal numeral twentieth (20th). This is a modification of the position assumed in (Appah

2019) where the semantic operator ORD was said to have scope over the cardinal numeral

constituent only. It is worth stressing that the ordinal reading in these constructions is possible

only when the transitive VP contains a cardinal numeral second constituent. As Appah (2019)

observes, we find VP constructions built around the verb tsia with non-numerals in

complement position elsewhere in the grammar of Akan. However, the relevant VPs do not

have the ordinal semantics at all. This is exemplified in (15), where the verb tsia with the noun

sika ‘money’ form a VP which means ‘to save money’.

(15) tsia sika

pile money

‘to save money’ (cf. Appah 2019: 8)

This fact supports the point made about the constructions that Christaller classifies as ordinal

numerals, but which are rejected as ordinals in this paper because they do not contain the

relevant distinguishing element – the cardinal numeral constituent. That is, there may be

constructions which look and/or function like numerals, but may not be classified as such

because they fail to meet the twin-expectation that they are numerals and that they are used to

assess quantitative properties of empirical objects (cf. Wiese 2003b, 2007). It is in this sense

that even the Akan construction for first does not qualify as a numeral. However, as noted

above, the idiosyncrasy of ordinal first appears to be a crosslinguistic feature (cf. Barbiers

2007; Booij 2009; Stolz & Veselinova 2013).

The second type of Akan ordinal numeral constructions have the structure tɔ do ‘lie in

position/come in at position …’ and a cardinal numeral that refers to a position in an ordered

set, as exemplified in (16).

(16) a. tɔ do enum

lie on five

‘lie in position (comes in at number) five (5th)’

b. tɔ do du

lie on ten

‘lie in position ten (10th)’

c. tɔ do dubiako

lie on eleven

‘lie in position eleven (11th).’

12

d. tɔ do ɔha

lie on hundred

‘lie in position hundred (100th).’

e. tɔ do eduanan ebien

lie on forty two

‘lie on position forty-two (42nd).’

f. tɔ do aha esia eduonum esia

lie on hundred six fifty six

‘lie in position six hundred and fifty-six (656th).’

g. tɔ do m-pem ebien na eduosuon awɔtwe

lie on PL-thousand two CONJ seventy eight

‘lie in position two thousand seventy-eight (2078th). (cf. Appah 2019: 8)

A construction which has one of these ordinal numerals occurring in it is found in the quotation

from the verse 13 of the first chapter of the second book of kings (2 Kings 1:13), as presented

in (17), from the Akuapem dialect.

(17) Na ɔ-san soma-a aduonum so safohene

and 3SG-return send-PAST fifty over captain

a ɔ-tɔ so abiɛsa ne n’-aduonum

REL. 3-fall on three CONJ 3POSS-fifty

‘Again, he sent a third captain of fifty with his fifty mean’ (NKJV)

As the data in (16) show, the numeral constituent may be simplex or complex. However, the

numeral constituent on its own does not express ordinality. It is only in conjunction with the

other elements of the construction that the ordinal meaning is expressed. Again, tɔ do/so in

these constructions has the structure of a typical VP. Therefore, as noted by Appah (2019),

with the addition of the numeral (e.g. tɔ do anan ‘4th’), the construction comes out as a kind of

ditransitive construction, with the structure [V NP Num]. There are alternative analyses of this

construction type (cf. Appah 2019: 11, n. 9) and a reviewer questions whether it is felicitous to

suggest that the resultantant construction is ditransitive. However, we believe that the choice

of terminology is supported by the observation that “[b]eside appearing as nominal modifiers,

numerals also occur independently of any nominal, as NPs in their own right … Such bare

numerals can occur as subjects and objects of sentences” (Hurford 1987: 158).

Working with the assumption that these ordinal constructions are ditransitive, we can

assume further that they inherit their structure from a ditransitive construction with the second

dependent slot pre-specified to be a cardinal numeral. Thus, because we know what the other

constituents of the construction are, we can posit a constructional idiom in which the first two

constituents (tɔ, do) are pre-specified, and the only variable slot is also specified to be a

numeral, as shown in (18).

(18) <[[[tɔ]i [do]j [Num]k]l]q ↔ [ORD [assume position NUM]l]q> (cf. Appah 2019)

13

Finally, it is worth pointing out that these constructions may replace the constructions from the

first set which occur in the names of the books of the Bible, as given in (11). Also, the Fante

translation of the quotation from 2 Kings1:13 uses ordinal numeral of the first kind presented

in (11), as shown in example (19).

(19) Na bio ɔ-soma-a eduonum do safohene a

and again 3-send-PAST fifty over captain REL.

ɔ-tsia ebiasa nye n’-eduonum

3-piles.to.make three CONJ 3POSS-fifty

‘Again, he sent a third captain of fifty with his fifty mean’ (NKJV)

5.1 Ordinal numerals and the partitioning of the ordinal space.

The foregoing discussion shows that ordinal numerals impose a two-way partition on the

ordinal space, as shown graphically in Figure 2. The first position (numbered 1 and shaded

black) is rendered as dzi kan or kan alone in appropriate context, as described above. All other

positions after first (numbered 2 and shaded grey) are referred to by means of ordinal numerals

which contain number words such as tsia ebien/tɔ do ebien ‘be second’. Here, because they

include numbers, they can refer to specific positions or ranks in the ordinal space, all of which

come subsequent to the first position. Last is excluded, as the final position in the series will

equally be identified by a specific numeral occurring in the relevant construction type – [tsia

NUM] or [tɔ do Num].

Figure 2: How ordinal numerals partition the ordinal space

6. The successor relation construction

The second means of expressing relative position within sets of objects is a class of VPs, which

we refer to as the successor relation constructions of succession constructions. These ordinal-

like VPs may be used to express the relative position of items in ordered series, much like

ordinal numerals. They were, therefore, previously classified as ordinal numerals (cf.

Christaller 1875). However, it is argued in this paper that they are not ordinal numerals for two

reasons. First, unlike ordinal numerals, they are not numerals and they do not contain actual

number words either and so they cannot be involved in number assignment. Secondly, the

subset of ordinal numerals discussed in (16), which are built around the verb tɔ, are constructed

on the basis of a subclass of the construction in this group by the addition of a cardinal numeral

like ebien ‘two’, as in tɔ do ebien ‘be second’. In the rest of this section, we discuss the structure

and distribution of these successor relation constructions and how they partition the ordinal

space.

1 2

(dzi) nkan

‘1st’

tsia ebien

tɔ do ebien

‘2nd’

tsia ebiasa

tɔ do ebiasa

‘3rd’

tsia anan

tɔ do anan

‘4th’

... ... ...

14

Three positions are distinguished, as far as the successor relation constructions are

concerned. These are first (rendered as front), last (rendered as back/hind end) and every other

position in-between. The first position is expressed as dzi kan (or just nkan, in appropriate

context), which is also the construction for ordinal first, as noted in (8) above and shown to

occur in the Akan translation of the name of the first book of Samuel in (9). For convenience,

those examples are repeated here as (20) and (21), respectively.

(20) dzi kan

occupy/assume front

‘to be first’ [lit. to occupy the front/to lead]

(21) Samuel nwoma a o-dzi kan no

Samuel book REL 3SG-assume front CFD

‘the first book of Samuel’ (lit. the book of Samuel which leads)

Using the successor relation constructions, the second and any subsequent position before the

final position in a set is expressed as dzi hɔ (22), tsia hɔ (23) or tɔ do (24). Because these

expressions do not contain any numeral, various locatives and relational nouns (cf. Osam et al.

2011) are employed to express the relative position of entities in ordered sets.

(22) dzi hɔ

occupy there

‘occupy the (next) place there/be next’

(23) tsia hɔ

pile there

‘be the next in the pile/be next’

(24) tɔ do

fall on

‘to follow (lit. fall on)/be next’

One effect of the absence of actual numeral constituents in these constructions, compared to

ordinal numerals, is that they do not refer to unique positions. Rather, they may be used for any

consecutive position relative to a given position within an ordered set. In this sense,

constructions (22), (23) and (24), which may be regarded as synonymous constructions, are all

ruled out of expressing the first position because, in principle, there has to be an entity/position

in place before another can be expected to follow. Thus, their semantics, which may be

rendered in general terms as ‘be next’ or ‘follow’, prevents them from being used to express

the first position in any set.

Using the successor relation constructions, the final position in a series is normally

expressed as either dzi ekyir ‘occupy the back part’ (25) or dzi ewiei ‘occupy the end’ (26).

(25) dzi ekyir

occupy/assume back/hind

‘to occupy the back-part/to be last’

15

(26) dzi ewiei4

occupy/assume end

‘to occupy the end/to be last’

It is worth pointing out that the construction in (25) may also be used to code consecutive

position, as in ‘the one that assumes the position behind/follows another’. Thus, it could be

used interchangeably with the constructions in (22) - (24) in appropriate contexts. In fact, these

constructions may be used for even the final position/item in a series, as if one had not come

to the end of the sequence. This will be the case, for instance, when calling a position “last” is

deemed inappropriate, politically incorrect, face-threatening, etc. For example, when

kindergarten children have finished a 20-meter race, the teacher handling them may avoid

saying a little child was “last” because it is potentially discouraging. Again, at a meeting where

the persons present were competing to raise funds for a project, the one announcing the result

avoided using the expression “last” in relation to the group that raised the least amount,

although the person used dzi kan ‘be first’, tɔ do ‘be next’, etc. The decision not to use the Akan

expression for last was a pragmatic one because it would have been face-threatening to use the

construction for last.

We have argued that the constructions in (22) - (24) and even (25) may be used for any

position after first and that this is due to the absence of number words (cardinal numeral)

constituents which would have restricted them to the expression of unique positions within the

set. Thus, the absence of a numeral in the construction correlates with the possibility of a

construction being used for any position after the first. The absence of actual numeral

constituents also means, as noted above, that the successor relations constructions are not

numerals and so they are not involved in number assignment. Indeed, where there is the need

to be specific, a cardinal numeral has to be introduced to form a proper ordinal numeral as

shown in (16). This is what was referred to in the introduction, where it was indicated that one

reason for arguing that successor relation constructions are not ordinal numerals, contra

previous characterization, is that actual ordinal numerals may be built on these ordinal-like

successor relation constructions. Hence, we can clearly distinguish them from Akan ordinal

numerals which must contain actual numbers, as discussed in Section 5, the only exception

being first, which generally tends to be different cross-linguistically.

Just like the ordinal numerals discussed above, the successor relations constructions

make use of structural options already available in the language system for their formal

expression. Thus, structurally, they are regular verb phrases, and so we may represent each of

them as a specific instantiation of the VP construction in Akan, as shown in (27).

(27) [V NP]VP

[[dzi]V [kan]NP]VP [[dzi]V [hɔ]NP]VP [[tsia]V [hɔ]NP]VP [[tɔ]V [do]NP]VP

[[dzi]V [ekyir]NP]VP [[twa]V [to]NP]VP

4 The noun ewiei is derived from the parasynthetic affixation of prefix a- and suffix -i to the verb wie ‘to finish’.

16

Again, as constructions, each expression is paired with a meaning specification, as

shown in (28), and each one means ‘be next’, which could be second, third, sixth, etc. They do

not refer to unique positions in the ordinal space.

(28) a. [[dzi]V [kan]NP]VP ‘be first’

b. [[dzi]V [hɔ]NP]VP ‘be next’

c. [[tsia]V [hɔ]NP]VP ‘be next’

d. [[tɔ]V [do]NP]VP ‘be next’

e. [[dzi]V [ekyir]NP]VP ‘be last’

f. [[twa]V [to]NP]VP ‘be last’

6.1 The successor relation constructions and the partitioning of the ordinal space

We observe, from the foregoing discussion, that successor relation constructions may be

grouped into two, depending on how they partition the ordinal space. However, there is a

common expression for ‘first’, dzi kan ‘occupy/assume front/to lead’. This is due to the fact

that constructions like dzi hɔ and tɔ do cannot be used to refer to an initial position in a set,

given their general meaning of addition to some already existing quantity/position. They refer

to consecutive positions and, on occasion, can be used even for the final position, creating the

impression that the reckoning continues. Thus, it would not be felicitous to refer to a first

position by an expression which presupposes an existing value/position.

The partition imposed by the successor relation constructions exemplified in (22) - (26)

may be represented graphically as Figure 3. The first subtype partitions the space into three,

as shown on the top row of Figure 3 – first position, rendered as dzi kan (numbered 1 and

shaded black), last position rendered as dzi ekyir/ewiei (numbered 3 and shaded ash), and others

in-between, rendered variously as dzi hɔ, tsia hɔ, or tɔ do (numbered 2 and shaded grey).

1 2 3

dzi kan

‘be first’

dzi hɔ/

tɔ do

‘be next’

dzi hɔ/

tɔ do

‘be next’

... dzi ekyi/

dzi ewiei

‘be last’

1 2

Figure 3: The partition of the ordinal space by successor relation constructions

The second subtype, as the bottom row of Figure 3 shows, imposes a two-way partition – first

position, rendered as dzi kan (numbered 1 and shaded black) and any subsequent position,

including the last in a non-finite ordered set, realised as tɔ do (numbered 2 and shaded grey).

As noted in the introduction, there is vehement disagreement (from a reviewer) over

the position taken in this paper that the constructions discussed in this section are not ordinal

numerals. It is argued, among others, that the position assumed in this paper amounts to

completely ruling out language variation for the encoding of one and the same concept. There

is no gain saying that I disagree with the objections raised. Indeed, there will be no need for

this paper, if I were to accept that even the successor relation constructions are all ordinal

numerals, because I would not be saying anything different from Christaller (1875). It has been

shown in the present paper that even within the same language there can be more than one

17

means of expressing ordinal meaning, as the discussion in section 5 shows. However, we have

also shown that not all the constructions so classified previously are ordinal numerals. This is

because we have clearly shown that, while they may refer to positions in ordered items, what

we call succession constructions or successor relation constructions are different in substance

from ordinal numerals because they may be the basis for the formation of a subclass of actual

ordinal numerals.

Another point that is canvassed in opposition to the claim that the constructions

discussed in this section are not ordinal numerals is that, like English second which derives

from Latin secundus ‘second’ which also derives from sequ-I ‘follow-INF’ which means to

follow (Veselinova 1997; Greenberg 2000:780), the verb phrases cited in this section could be

seen as early variants of evolving alternative expressions for ‘second’. Whereas this sounds

interestingly plausible, there is no evidence that the constructs are lexicalizing or

grammaticalizing into ordinal numerals meaning second. Additionally, they are not used for

second position only. As indicated above, any of the constructs discussed in this section could

refer to the second, third, sixth, seventh, tenth, or even hundredth. What is required is that there

exists some item or position, which could be any position at all, so that any subsequent one can

dzi hɔ or tɔ do ‘be next to it’, .

7. Conclusion

In this paper, we have discussed the means by which positions in ordered sets are expressed in

Akan. We started with Christaller’s classes of “Ordinal numerals” and went on to show that

there is basis for distinguishing two kinds of constructions from the constructs reported in

Christaller (1875), working with the idea of number assignment which has to do with using

numbers to assess the properties of empirical objects. The first class has constructions that

express rank or position of items (first, second, third, etc.) within ordered sequences

numerically (Stump 2010). We have argued that they are actual ordinal numerals per the

criterion of number assignment. In these constructions, the numeral makes it possible to refer

to uniques positions in the ordinal space. The second class has constructions that express

successor relations non-numerically and so cannot refer to unique positions or ranks in ordered

sets. They simply mean ‘be next’ or follow. We have argued that they are not ordinal numerals.

Thus, the former can be differentiated from the latter because the former contain cardinal

numeral constituents. Additionally, constructions from the latter group may be the base for the

formation of actual ordinal numerals (Appah 2019). The distinction made in this paper makes

Akan, to some extent like English which, in addition to ordinal numerals like fourth, fifth, etc.,

has items like next, follow, etc., which express succession. No one ever calls next an ordinal

numeral because it is not a numeral and it does not participate in number assignment. The

obvious difference is that English next is an adjective, but Akan Uses a VP which is part of a

relative clause.

In the discussion, we showed that all the constructions have the structure of VPs. Thus,

following Appah (2019) we argued that the constructions inherit their formal structure from

already existing VP constructions in the language. We indicated that the two groups of

constructions partition the ordinal space differently and we illustrated the different partitions

graphically. We also indicated that the partitioning of the ordinal space may be influenced by

pragmatic factors. For example, a potential three-way partition may be reduced to a two-way

partition, if the speaker believes that it may be politically incorrect or potentially face

18

threatening to refer to the final position in a set by the construction for last. In that case, the

expression for next would be used, as if the speaker had not come to the end of the reckoning.

Acknowledgements

I would like to thank an anonymous reviewer of Constructions+ whose suggestion of a need

for a single focus in the review of Appah (2019) became the motivation for the present paper.

I am most grateful. I am also grateful to the editors and two anonymous reviewers of SKASE

JTL for their comments that helped to improve this paper, although I have been a bit stubborn

in not accepting all their recommendations. Thus, I am solely responsible for any remaining

shortcomings of the present paper.

Abbreviations

1 First person

3 Third person

CONJ Conjunction

COMP Complement

CxM Construction Morphology

CFD Clause Final Determiner

HAB Habitual

Lit. Literal meaning

N Noun

NKJV New King James Version

NP Noun Phrase

NUM Numeral/numerical value

ORD Ordinal

PL Plural

REL Relative marker

SG Singular

SEM Semantics

V Verb

VP Verb phrase

References

Appah, Clement K. I. 2013. Construction Morphology: Issues in Akan Complex Nominal Morphology.,

Lancaster: Lancaster University. (PhD dissertation).

Appah, Clement K. I. 2015. On the syntactic category of Akan compounds: A product-oriented

perspective. Acta Linguistica Hungarica 62(4): 361-394. doi: 10.1556/064.2015.62.4.1

Appah, Clement K. I. 2017. On holistic properties of morphological constructions: the case of Akan

verb-verb nominal compounds. Acta Linguistica Hafniensia 49(1): 12-36. doi:

10.1080/03740463.2016.1242331

19

Appah, Clement K. I. 2019. Ordinal numeral constructions in Akan. Constructuions 1/2019: 12pp.

Barbiers, Sjef. 2007. Indefinite numerals ONE and MANY and the cause of ordinal suppletion. Lingua

117(5): 859-880.

Booij, Geert E. 2002. Constructional Idioms, Morphology, and the Dutch Lexicon. Journal of Germanic

Linguistics 14(4): 301-329. doi:10.1017/S1470542702000168

Booij, Geert E. 2007. Construction Morphology and the Lexicon. In F. Montermini, Boyé, G. &

Hathout, N. (Eds.), Selected Proceedings of the 5th Décembrettes: Morphology in Toulouse

(pp. 34-44). Toulouse: Cascadilla.

Booij, Geert E. 2009. Constructions and lexical units: an analysis of Dutch numerals. Linguistische

Berichte Sonderheft 19: 1-14.

Booij, Geert E. 2010a. Construction morphology. Language and Linguistics Compass 4(7): 543-555.

Booij, Geert E. 2010b. Construction morphology. Oxford: Oxford University Press.

Booij, Geert E. 2012. Construction Morphology, a brief introduction. Morphology 22(3): 343-346.

Brainerd, Barron. 1966. Grammars for Number Names. Foundations of language 2(2): 109-133.

Brainerd, Barron. 1968. On the Syntax of Certain Classes of Numerical Expression. In H. Brandt

Corstius (Ed.), Grammars for number names (pp. 9-40). Dordrecht-Holland: D Reidel

Publishing Company.

Brandt Corstius, Hugo. 1968. Grammars for number names (Vol. 7). Dordrecht: D Reidel Publishing

Company.

Broohm, Obed N. 2019. Issues in Esahie Nominal Morphology: From Inflection to Word Formation.

PhD, Department of Culture and Civilization, University of Verona, Verona.

Christaller, Johannes G. 1875. A Grammar of the Asante and Fante Language called Tshi (Chwee, Twi)

based on the Akuapem dialect with reference to the other (Akan and Fante) dialects. Basel:

Basel Evangelical Missionary Society.

Corbett, Greville G. 1978. Universals in the syntax of cardinal numerals. Lingua 46(4): 355-368. doi:

10.1016/0024-3841(78)90042-6

Dehaene, Stanislas. 1997. The Number Sense. New York: Oxford University Press.

Dehaene, Stanislas. 2001a. Author’s Response: Is Number Sense a Patchwork? Mind & Language

16(1): 89-100. doi: 10.1111/1468-0017.00159

Dehaene, Stanislas. 2001b. Précis of The Number Sense. Mind & Language 16(1): 16-36. doi:

10.1111/1468-0017.00154

De Wit, Astrid. 2018. The Semantics of the Simple Tenses and Full-Verb Inversion in English: A Story

of Shared Epistemic Schemas. Constructions and Frames 10(2): 210-33. doi:

https://doi.org/10.1075/cf.00019.wit.

20

Dolphyne, Florence. A. 1988. The Akan (Twi-Fante) Language: Its Sound Systems and Tonal Structure.

Accra: Ghana Universities Press.

Dugas, Edwige 2018. Form/Meaning Asymmetry in Word Formation: The Case of Non-Nouns in

French. Constructions and Frames 10(2): 178-209. doi: https://doi.org/10.1075/cf.00018.dug.

Fillmore, Charles J., Kay, Paul, Michaelis, Laura A., & Sag, Ivan A. 2003. Construction Grammar.

Stanford and Chicago: CSLI Publications and University of Chicago Press.

Goldberg, Adele E. 1995. Constructions: a construction grammar approach to argument structure.

Chicago: University of Chicago Press.

Goldberg, Adele E. 2006. Constructions at work: the nature of generalization in language. Oxford:

Oxford University Press.

Goldberg, Adele. E. & Johan. van der Auwera. 2012. This Is to Count as a Construction. Folia

Linguistica 46(1): 109-32

Greenberg, Joseph H. 2000. Numeral. In Geert Booij, Christian Lehmann & Joachim Mugdan in

collaboration with Wolfgang Kesselheim & Stavros Skopeteas (eds.), Morphology: An

international handbook on inflection and word-formation. (pp.770–783). Berlin: Mouton de

Gruyter.

Hurford, James. R. 1987. Language and Number: The Emergence of a Cognitive System. Oxford: Basil

Blackwell.

Hurford, James. R. 2001. ‘Numeral Systems’. in, N. J. Smelser & P. B. Baltes (eds.) International

Encyclopedia of the Social and Behavioral Sciences (pp. 10756-10761). Amsterdam:

Pergamon.

Jackendoff, Ray S. 1997. The architecture of the language faculty. Cambridge, MA: MIT Press.

Jackendoff, Ray S. 2002. Foundations of language: brain, meaning, grammar, evolution. Oxford/New

York: Oxford University Press.

Jackendoff, Ray S. 2008. Construction after construction and its theoretical challenges. Language 84(1):

8-28.

Jurafsky, Daniel. 1992. An on-line computational model of sentence interpretation. University of

California, 1992. Report No. UCB/CSD 92/676, Berkeley. (PhD Dissertation).

Lawer, Richard A. 2017. Compounding in Dangme. Department of Linguistics, University of Ghana,

Accra. (MPhil thesis).

Mel'čuk, Igor. 1994. Suppletion: Toward a Logical Analysis of the Concept. Studies in Language 18(2):

339-410.

Michaelis, Laura A., & Lambrecht, Knud. 1996. Toward a Construction-Based Theory of Language

Function: The Case of Nominal Extraposition. Language 72(2): 215-247.

21

Osam, E. Kweku, Duah, Reginald A., & Blay, Afua M. 2011. The so-called postpositions in Akan: A

reconsideration. Journal of West African Languages XXXVIII(2): 107-118.

Saah, Kofi K. (2010). Relative clauses in Akan. In E. O. Aboh & J. Essegbey (Eds.), Topics in Kwa

Syntax. (pp. 91-107). Dordrecht: Springer.

Stampe, David. 1976. Cardinal Number Systems. In S. S. Mufwene, Walker, C. A. & Steever, S. B.

(eds.), Papers from the Twelfth Regional Meeting, Chicago Linguistics Soceity (pp. 594-609).

Chicago, Il.: Chicago Linguistics Soceity.

Stolz, Thomas, & Veselinova, Ljuba N. 2013. Ordinal Numerals. In M. S. Dryer & Haspelmath, M.

(Eds.), The World Atlas of Language Structures Online. Munich: Max Planck Digital Library.

Stump, Gregory T. 2010. The derivation of compound ordinal numerals: Implications for morphological

theory. Word Structure 3(2): 205-233. doi: doi:10.3366/word.2010.0005

van Katwijk, A. 1965. A Grammar of Dutch Number Names. Foundations of language 1(1): 51-58.

Veselinova, Ljuba. 1997. ‘Suppletion in the Derivation of Ordinal Numerals: A Case Study’. in B.

Bruening (ed.) Proceedings of the Eighth Student Conference in Linguistics (Scil-8), (pp. 71-

92). Cambridge, MA: MITWPL.

von Mengden, Ferdinand. 2010. Cardinal Numerals: Old English from a Cross-Linguistic Perspective.

Berlin: Mouton De Gruyter.

Wiese, Heike. 2003a. Iconic and non-iconic stages in number development: the role of language. Trends

in Cognitive Sciences 7(9): 385-390.

Wiese, Heike. 2003b. Numbers, language, and the human mind. Cambridge Cambridge University

Press.

Wiese, Heike. 2007. The co-evolution of number concepts and counting words. Lingua 117(5): 758-

772.

Clement Kwamina Insaidoo Appah

Department of Linguistics,

University of Ghana

P.O. Box LG61

Legon-Accra,

Ghana

cappah@ug.edu.gh

In SKASE Journal of Theoretical Linguistics [online]. 2019, vol. 16, no. 4[cit. 2019-12-20].Available

on web page http://www.skase.sk/Volumes/JTL42/pdf_doc/01.pdf. ISSN 1336-782X